Fiber Bragg grating-based sensor for monitoring ... · Fiber Bragg grating-based sensor for monitoring respiration and heart activity during magnetic resonance imaging examinations

Fiber Bragg grating-based sensor formonitoring respiration and heart activityduring magnetic resonance imagingexaminations

Łukasz DziudaFranciszek W. SkibniewskiMariusz KrejPaulina M. Baran

Downloaded From: https://www.spiedigitallibrary.org/journals/Journal-of-Biomedical-Optics on 03 Apr 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use

Fiber Bragg grating-based sensor for monitoringrespiration and heart activity during magneticresonance imaging examinations

Łukasz Dziuda,a Franciszek W. Skibniewski,a Mariusz Krej,b and Paulina M. Baranb

aMilitary Institute of Aviation Medicine, Technical Department of Aeromedical Research and Flight Simulators, ul. Zygmunta Krasińskiego 54/56, 01-755 Warszawa, PolandbMilitary Institute of Aviation Medicine, Department of Aviation Bioengineering, ul. Zygmunta Krasińskiego 54/56, 01-755 Warszawa, Poland

Abstract. We present a fiber-optic sensor for monitoring respiration and heart activity designed to operate in themagnetic resonance imaging (MRI) environment. The sensor employs a Plexiglas springboard, which convertsmovements of the patient’s body lying on the board (i.e., lung- and heart-induced vibrations) to strain, where afiber Bragg grating attached to the board is used to measure this strain. Experimental studies are carried out duringthoracic spine MRI examinations. The presence of the metal-free sensor construction in the MRI environment doesnot pose a threat to the patient and has no influence over the quality of imaging, and the signal is identical to thatobtained without any electromagnetic interference. The results show that the sensor is able to accurately reflect theballistocardiographic signal, enabling determinations of the respiration rate (RR) and heart rate (HR). The data deliv-ered by the sensor are normally distributed on the Bland–Altman plot for the characteristic point determination andexhibit clear dependence on the RR and HR values for the RR and HR determinations, respectively. Measurementaccuracies are better than 7% of the average values, and thus, with further development, the sensor will be imple-mented in routine MRI examinations. © The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License.

Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI. [DOI: 10.1117/1

.JBO.18.5.057006]

Keywords: ballistocardiographic signal; fiber Bragg grating sensors; heart rate; hyperventilation; magnetic resonance imaging examina-tions; respiration rate.

Paper 130034R received Jan. 21, 2013; revised manuscript received Apr. 5, 2013; accepted for publication Apr. 22, 2013; publishedonline May 22, 2013.

1 IntroductionRecently, magnetic resonance imaging (MRI) has become oneof the most common and reliable diagnostic methods in medi-cine. The results of the still improving imaging performances,new functionalities, and more affordable prices represent a pro-nounced trend toward equipping hospitals and clinics with MRIscanners.1 The rapid development of this technology has createdthe need to construct devices capable of monitoring vital signsof at-risk patients during MRI examinations. The risk groupin the context of MRI includes neonatal, pediatric, disabled,anesthetized, and insensible patients.2 The monitoring of lifefunctions is also recommended for persons with implanted pace-makers, patients developing reactions to contrast media, andpersons with mental disorders who may not be able to commu-nicate with the system operator or to use the alarm button.However, efforts are being made to monitor at least somebasic physiological parameters such as respiration and heartactivity3 during each MRI examination, regardless of the patientcategory.

Although MRI is a noninvasive method of investigation anddoes not expose patients to ionizing radiation, it has somepsychological effects on patients. Some researchers4,5 havealready demonstrated that 5% to 10% of patients undergoing

MRI examinations experience severe claustrophobia or panicattacks during the scanning process, whereas 30% report milderdistress and fear. Such increased stress stimuli and arousal mayin turn lead to a phenomenon known as hyperventilation, whichis increased ventilation of the lungs and significant accelerationand deepening of the breath, often resulting in fainting.

Hyperventilation has numerous theoretical and empiricallinks to anxiety, panic, and fear episodes.6,7 This phenomenonwas investigated in order to understand psychological andphysiological mechanisms connected with hyperventilation thatcause and maintain anxiety. On the other hand, attempts werealso made to find a link between hyperventilation and the thera-pies used to treat different types of anxiety disorders. Someresults8 support the idea of higher variability and irregularity inthe respiratory patterns of subjects with panic disorder. Otherclinical studies9 have shown that controlling and normalizingthe subject’s breathing activity helps to reduce both anxietyand panic symptoms. On the basis of this assumption, differenttraining and therapeutic programs10 have been created and car-ried out with the aim of teaching patients to breathe in a specificway to avoid or reverse hyperventilation, e.g., capnometry-assisted respiratory training.11 Furthermore, attempts have alsobeen made to study the relationship between different anxietydisorders and cardiological parameters. The latest literature12

reveals an increased heart rate (HR) level in a group of peoplesuffering from panic, anxiety disorders, and hyperventilationsymptoms.

In light of recent scientific reports, it seems obvious thatMRI examinations can be a source of extreme stress and

Address all correspondence to: Łukasz Dziuda, Military Institute of AviationMedicine, Technical Department of Aeromedical Research and FlightSimulators, ul. Zygmunta Krasińskiego 54/56, 01-755 Warszawa, Poland. Tel:+48 22 685 2621; Fax: +48 22 685 2715; E-mail: [email protected]

Journal of Biomedical Optics 057006-1 May 2013 • Vol. 18(5)

Journal of Biomedical Optics 18(5), 057006 (May 2013)


http://dx.doi.org/10.1117/1.JBO.18.5.057006






cause anxiety and panic symptoms in some patients. People whoexperience anxiety, fear, or hyperventilation during an MRI scanmay either discontinue the procedure altogether or refuse toundergo MRI surveys in the future. In practice, it is importantto identify those who are likely to panic and experience anxiety-related symptoms in the MRI unit as early as possible becauseonly early detection ensures that appropriate actions can betaken to avoid psychological discomfort for the patient under-going the scan and unnecessary use of the machine, as well asstaff involvement in the entire procedure. Real-time monitoringof certain life functions in the patient, such as respiration rate(RR) and HR, can be helpful in controlling the arousal and anxi-ety level associated with the MRI examination itself and predict-ing the probability of a panic attack.

Conventional monitoring equipment is not designed to workin the harsh MRI environment, distinguished by a magneticinduction of 1.5, 3.0 T, or even higher values and with radiofrequency impulses of considerable electromagnetic (EM)field intensity.13 The presence of metal or electronic componentsunder such conditions can cause burns on the patient’s skin andalter the EM field distribution and, thus, degrade the quality ofimaging. Additionally, the signal transmitted using electricalconductive wires is vulnerable to the effects of a strong EMfield, which complements the list of drawbacks of typical elec-tronic measuring devices. On the other hand, optical sensors arefully penetrable in terms of the EM radiation applied in MRI,causing no interference. They are free frommetal parts and func-tion via optical fibers, resulting in their immunity to EM fields.

There are solutions based on phase interferometry, in whichthe signal directed toward the sensing part of the optical fiberthat is situated close to the patient is compared with the outgoingsignal that is modified by confounding factors such as respira-tion and heartbeat.14–19 The lung- and heart-induced body move-ments cause phase changes between the input and output signalsand further generate a time-varying serial pattern of fringes. Amajor disadvantage of this measuring method is the necessity ofusing sophisticated signal analysis in order to detect breathingand cardiac activity. A similar problem concerns sensors in themodalmetric configuration, which are comprised of multimodefibers sensitive to mechanical strain so that acousto-mechanicalsignals from the patient are coupled into the single-mode opticalfiber.20 The modal field distribution in the multimode fiberchanges in time proportionally to body movements and isdetected by the single-mode fiber. Another problem associatedwith the modalmetric sensors is the difficulty in obtaining simul-taneous monitoring of respiratory and heart function.

We previously demonstrated that it is possible to detect res-piration and heart activity simultaneously, using a fiber Bragggrating (FBG) mounted inside a pneumatic cushion, to be placedbetween the backrest of the seat and the back of the monitoredperson.21–23 Other pillow-based solutions have been reported inthe literature,24–26 in which a microbending fiber-optic sensorwas used instead of an FBG. In this paper, we show a sensorof a simplified construction, designed to be immune to theEM field, and thus suitable for MRI applications. Similar tech-niques for breath detection have been demonstrated by othergroups, but their sensors are embedded into special textilebelts or vests to be fixed to the monitored person’s body.27–39

The sensor proposed in this paper is configured for measure-ments that do not require any special wearable textiles, therebyavoiding the need to carry out processes for preparing the patientfor monitoring. The solution closest to the proposed sensor is

one comprising a 12-element FBG array embedded into abed,40 although only laboratory trials evaluating the RR detec-tion have been carried out so far. This article describes in detailthe issue of patient monitoring during MRI scanning using theFBG-based sensor and discusses the proposed solution withregard to other known devices.

2 Methods

2.1 Fiber Bragg Gratings

The sensor is designed using a Bragg grating inscribed into asingle-mode optical fiber as the sensing element. FBGs havebeen one of the most widely researched optical componentswithin the last decade. They have found manifold applicationsin telecommunications, laser, and sensor technology. The sensorapplications of FBGs alone cover many fields, such as structuralhealth monitoring,41 monitoring of conditions in electricalplants,42 and physiological activity monitoring.43 FBGs owetheir popularity to their immunity to optical signal intensitymodulation, which is possible on account of their spectralencoding as well as excellent multiplexing and self-referencingcapabilities, thus allowing for the addressing of several sensorswith one optical cable. At the same time, they offer the standardbenefits of fiber-optic technology such as EM immunity,intrinsically safe modes of operation, a chemically inert nature,small size, and light weight.

The FBG acts as a stop-band filter, except that part ofthe incident light, Iin, is not stopped but is reflected fromperiodically arranged layers formed by refractive index modu-lation (see Fig. 1). The reflected beams from each of the layersdestructively interfere with each other unless they are all inphase. This only occurs at one wavelength, known as theBragg wavelength (or the central wavelength), λB, which isgiven by

λB ¼ 2neffΛ; (1)

where neff is the effective refractive index of the fiber core,44 and

Λ is the period of the refractive index modulation.The light reflected from the grating, Ire, assumes the spectral

form of a sharp “peak” located at the Bragg wavelength,whereas the transmitted light, Itr, is characterized by a narrow“dip” at the same wavelength. In most sensing applications, thecentral wavelength is determined on the basis of the peak posi-tion, as the reflected signal shape can be more easily demodu-lated. Thus, FBG-based sensors offer the unique advantages of asingle-ended connection to the interrogation system.

2.2 Sensor Construction

A starting point from which to build the sensor for monitoringrespiration and heart activity was the susceptibility of the FBGto strain, which produces a shift in the Bragg wavelength, ΔλB,and that is described by the following relation:41

ΔλBλB

¼�1 −

n2eff2

½p12 − νðp11 þ p12Þ��Δε ¼ ð1 − ρeÞΔε;

(2)

where p11 and p12 are the silica photo-elastic tensor compo-nents, ν is the Poisson’s ratio, Δε is the strain change, andρe is the stress-optic coefficient. For a germanium-doped


Dziuda et al.: Fiber Bragg grating-based sensor for monitoring respiration and heart activity. . .


fiber core and a wavelength of 1550 nm, the strain change can besummarized as follows:23

Δε ¼ ΔλB1.212

: (3)

The sensor design is based on the approach in which aspringboard is used to convert body movements, includinglung- and heart-induced vibrations, to strain, and an FBGattached to the board is used to measure this strain. The instanta-neous spectral position of the FBG reflection peak can be cali-brated in terms of respiration and heart activity readings. For thesensor to work properly, the board has to be placed between thebed mattress and the back of the patient.

As the base of the sensor, a polymethyl methacrylate(PMMA) (i.e., Plexiglas) board with dimensions of 220×95 mm2 and a thickness of 1.5 mm was used. A springboard

of this size and thickness provides proximity to the body withouta loss of comfort for the patient lying on it. At the same time, theboard size ensures both that the body vibrations are sufficientlytransferred to the FBG and that the vibrations can be decoded inthe form of FBG elongations/contractions. The FBG, with λB ¼1541 nm at a temperature of 20°C, was attached with epoxyadhesive to the center part of the board, as shown in Fig. 2.After adhesive curing, the peak position shifted in the directionof higher wavelength values; its full width at half maximumincreased slightly, and the peak amplitude decreased. One ofthe FBG pigtails was spliced with a pigtail that ended withan angled physical contact fiber connector connector capableof connecting the interrogation system. The splice protectorwas equipped with a ceramic reinforcing rod, instead of asteel one, in order to allow the sensor to be used in an MRI envi-ronment. The excess fiber between the FBG and the spliceprotector was formed into a loop, consisting of several turns,

Fig. 1 Principle of operation of an FBG.

Fig. 2 Construction of the sensor.




and was taped to the PMMA board. The other FBG pigtail wascut off a few centimeters from the grating and also taped tothe board.

2.3 Experimental Setup

A block diagram of the experimental setup is depicted in Fig. 3.The study was conducted at the Military Institute of AviationMedicine, Warsaw, Poland, equipped with the Achieva 1.5 TMRI scanner by Philips, Andover. The sensor was placed inthe MRI examination room between the mattress of the tracktable in the MRI system and the back of the lying patient inthe area of the shoulders to ensure proximity to the heart andthe lungs. The sensor lead, in the form of a 10-m opticalcable, was threaded through an opening in the wall and con-nected to the FBG interrogation system in the MRI system oper-ator’s room. To interrogate the sensor, a commercially availablesm130-700 unit from Micron Optics, Atlanta was used.45 Thedevice consists of two main modules: a broadband light sourceto illuminate the FBG, and a high-speed scanning Fabry–Pérot(FP) tunable filter to analyze the signal reflected from the gra-ting. Using a FP tunable filter is the most common technique formonitoring an FBG; any changes in the Bragg wavelengthlocated within the free spectral range of the FP filter can betracked by the spectrally scanning filter.23 The sm130-700 inter-rogator allows for scanning a band of 1510 to 1590 nm with afrequency of 1 kHz and a resolution of 1 pmpk−pk, which in turn,according to Eq. (3), ensures measurement of strains as low as0.8 μεpk−pk. The data from the interrogator were sent via theEthernet interface to a laptop PC with software developed forsignal visualization, data archiving, processing, and analysis.22

The experimental setup was equipped with respiration andelectrocardiogram (ECG) recorders from Invivo, Gainesville,46

designed to operate in the MRI environment, as a source of sig-nals used to determine the reference RR and HR. The respirationand ECG signals were acquired from the patient using pneu-matic bellows placed on the patient’s chest and carbon electro-des fixed to his or her chest, respectively, and transmitted fromthe recorders to a PC wirelessly via a 2.4 GHz radio link. As thedata were transmitted in two different ways (i.e., by Ethernet andradio), to minimize the possible occurrence of a shift between

the signals, they were synchronized by means of motion artifactsintroduced in the reference signals and the sensor signal at thebeginning and the end of the recording.

2.4 Subjects

Twelve adult volunteers were involved in the experimental stud-ies (four females, eight males); age: 33.25� 6.06 (mean� SD)years, weight: 78.42� 17.43 kg, and height: 177.17� 8.54 cm .When the sensor was placed under the back of each patient inthe same manner, no significant correlations were observedbetween the quality of the received signal and the age, weight,or height of the subjects. Any differences in the representation ofthe sensor signal are instead a result of individual features of thesubjects, such as the size of internal organs or the pattern ofblood vessels.

3 Results

3.1 Frequency Response

For the purpose of demonstrating the system’s frequencyresponse, the sensor was vibrated by a modified high-powerloudspeaker driven by a signal generator, as shown in Fig. 4.A reference FBG was mounted on the flexible cone suspensionof the loudspeaker to provide a strain measurement that is pro-portional to the cone excursion. This allowed the sinusoidalvibration to be set to a constant amplitude within the measure-ment range. The signal amplitude recorded by the sensor thatwas referred to the signal amplitude measured by the referenceFBG is shown in Fig. 5 in the form of transmittance correspond-ing with the frequency response of the measurement system.The sensor performance within the range of the ballistocardio-graphic (BCG) signal being measured (i.e., 1 to 20 Hz17,47)shows that signals are sensed without any degradation of thephysiological information. The BCG frequency range liesclearly below the resonant frequency of the sensor.

3.2 Determining RR

The main phase of the experiments covered clinical evaluationof the sensor’s RR and HR measurement capabilities, in real

Fig. 3 Experimental setup.




MRI conditions. Each of the subjects was asked to lie down onthe track table of the MRI scanner, and the physiological activ-ities of their bodies were recorded during a 5-min exposure thatwas part of a thoracic spine examination. Thus, the total record-ing time was 60 min. A 14-s portion derived from the entirerecording made on a 36-year-old man (weight: 86 kg; height:183 cm) is shown in Fig. 6. The upper trace represents the res-piration curve recorded with the reference module for

respiration monitoring, and the lower one shows the correspond-ing part of the signal coming from the sensor (shown in gray).Additionally, the original signal was denoised through an aver-aging of 250 samples (with the result shown by the black line).23

The slowly changing “waving” pattern in the sensor signalreflects the activity of the lungs. The arrows indicate the maxi-mum values of the signal during subsequent breaths used by theauthors as distinctive points for determining an RR trace. In this

Fig. 4 Setup for determining the system’s frequency response.

Fig. 5 Frequency response of the measurement system.

Fig. 6 14-s portion of a recording with visible respiration wave.




way, n local maxima were extracted for the processed 5-minsensor signal. The RR value at the time tn, at which the n’thmaximum occurred, is defined by the formula

RRn ¼60

tn − tn−1: (4)

Figure 7 shows the 5-min RR traces obtained for each of thesubjects, successively, using both (a) readings from the refer-ence recorder and (b) the set of local maxima extracted fromthe sensor signal. In both cases, the reference RR trace andthat obtained on the basis of the sensor signal are of similarshape, and therefore the Bland–Altman plot was used for quan-titatively assessing the sensor’s RR measurement capability.48,49

In this graphical method, the differences between the two tech-niques (i.e., x1 − x2) are plotted against the averages of the twotechniques [i.e., ðx1 þ x2Þ∕2]. The better the reproducibility isconsidered to be, the narrower the �1.96 SD range containingapproximately 95% of the results is. The accuracy of determin-ing the characteristic points from the sensor signal in relation tothose determined from the reference signal is shown in Fig. 8(a)for the total pool of data (all subjects). Among the values,96.10% lie within the limits of agreement (LoA); the resultsshow no systematic error, have no proportional character, anddo not depend on the frequency of the distinctive point occur-rence. Going further, the sensor’s RR determination capability isestimated in Fig. 8(b), in which 95.02% of the values lie withinthe LoA of 0.89 respirations per minute (rpm). In contrast to theBland–Altman assessment of the characteristic point determina-tion, the sample distributions for the RR determination clearlyshow the dependence on the RR values. This can be explainedby the fact that the error in determining characteristic points thatis independent on the RR translates into the error in determiningthe RR, which increases with RR, according to Eq. (4).

The sensor’s RR measurement capability is also estimated inFig. 9 by calculating the differences between the reference andmeasured RR values in relation to the average value betweenthem. The relative error span is below 7% [mean relativeerror (MRE) of RR detection: 5.24%� 1.02%], which in rela-tion to the number of inhalations per minute gives a value of

approximately �1.2 rpm [mean absolute error (MAE) of inter-vals between respiration events: 0.65%� 0.30%]. The results ofthe sensor’s RR measurement capability evaluation are summa-rized in Table 1 for each of the subjects as well as for the totalpool of data.

3.3 Determining Heart Rate

The cyclic disturbances seen on the envelope of the sensor signalshown earlier in Fig. 6 represent the activity of the heart. Hencethe following stage of experimental studies included sensor sig-nal processing in the context of the sensor’s HR measurementcapability. For a better visualization of the heart activity and dueto the need for HR detection, the sensor signal was normalizedby reducing it to the reference level of zero through low-passfiltering with a cutoff frequency of 60 Hz, which did notcause a loss of physiological information. Figure 10 showsan enlarged 2.5-s portion of a recording made with the samemale subject whose respiration traces were shown in the pre-vious section. The upper trace represents the reference ECG sig-nal (CM5 lead), whereas the lower one shows the correspondingpart of the signal recorded by the sensor (gray line). In additionto the low-pass filtering, the signal was denoised by averagingover 25 samples (black line).23

According to the literature,50–53 the general waveform of theBCG consists of an early peak (H, Fig. 10) related to the motionof the heart early in systole, and the main IJK complex related tothe ventricular ejection and aortic flow. It can be assumedapproximately that the highest peak (J) occurs at the maximumvalue of the ECG T-wave. It is easy to see that the BCG signalacquired by the sensor is also a sort of a kinetocardiogram54,55 inwhich the movement of the subject’s back in relation to the mat-tress of the track table is registered rather than the movement ofthe chest. However, the signal shape can also reflect the flow ofblood in the organs of the body.

The arrows shown in Fig. 10 indicate the J peaks of the BCGsignal during subsequent heartbeats used for determining theHR trace, in a manner similar to the RR detection procedure.Thus,m J peaks (local maxima) were extracted for the processed5-min signal from the sensor. The HR value at the time tm, atwhich the m’th maximum occurred, is defined as follows:

Fig. 7 Respiration rate extracted from (a) reference signal and (b) sensor signal shown successively for all of the subjects.




HRm ¼ 60

tm − tm−1: (5)

Figure 11 shows 5-min HR traces obtained for each of thesubjects using both (a) the ECG recorder and (b) the HR detec-tion procedure performed on the sensor signal. Again, the sensorand reference traces have similar shapes, and the differencesbetween them are better depicted in Fig. 12 by means of theBland–Altman plot. First, the accuracy of determining the Jpeaks from the BCG signal in relation to the R waves deter-mined from the ECG signal is shown in Fig. 12(a) for thetotal pool of data. 95.89% of the considered samples lie withinthe �1.96 SD range. Subsequently, the sensor’s HR determina-tion capability is estimated in Fig. 12(b). 95.11% of the valueslie within the LoA, and this range is equal to 3.64 beats per

minute (bpm). In the same way as the respiration data shownpreviously, the data for the distinctive point determination arerelatively normally distributed (i.e., there are no systematic orproportional errors), and no dependence on the frequency ofthe distinctive point occurrence is observed, whereas the HRsample collections strongly depend on the HR value. Again,the error in determining characteristic points that is independenton the HR translates into the error in determining the HR (wherethe error increases with HR), according to Eq. (5).

In addition to the Bland–Altman analysis, a fractional errorwas shown in Fig. 13 to characterize the usefulness of the pro-posed measurement method. The error varies within the range of6.5% (MRE of HR detection: 5.71%� 0.61%), which corre-sponds with approximately�3 bpm (MAE of beat-to-beat inter-vals: 2.40%� 0.46%). Table 2 summarizes the results of the

Fig. 8 Determination of (a) distinctive points and (b) RR assessed using the Bland–Altman method.

Fig. 9 Relative error of RR determination as a fraction of the average value.




sensor’s HR measurement capability evaluation, both for indi-vidual subjects and for the total pool of data.

3.4 Patient Comfort and Safety

At the end of each test, the subject was asked about any negativefeelings caused by the presence of the sensor in his or her back,such as excessive rigidity or size of the board, leading to galls orbruises on the subject. None of the subjects reported any symp-toms of discomfort caused by the sensor location; no one suf-fered any prejudice. Finally, the same male subject, whoserespiratory and cardiac recordings are shown above in Figs. 6and 10, was asked to undergo a complete thoracic spine MRIexamination for the purpose of assessing the comfort of a sub-ject remaining in the same position for a long time with his backin contact with the sensor. Figure 14 shows (a) RR and (b) HRtraces obtained on the basis of the sensor signal acquired duringa 39-min exposure. The subject does not need to lie down at adesignated position to enable the sensor to acquire his

physiological signals. He was able to simply remain in thesupine position identical to that taken by patients during anormal spine MRI examination. The physiological signals werecorrectly acquired from the subject during the entire duration ofthe scan. The subject did not complain of any discomfort asso-ciated with the presence of the sensor in his back.

High values of the EM field generated during the examina-tion did not affect the optical signal propagated in the fiber anddid not disturb the sensor operation even once. The clinicalevaluation proved that the signal was identical to that obtainedwithout exposure to the EM field. Moreover, the presence of thesensor during the MRI examination did not pose a threat to thepatient and had no influence over the quality of the imaging.Figure 15 shows an image of the region of the body underwhich the sensor was placed; the proposed transducer is trans-parent to the MRI system. It was verified that the sensor does notintroduce any artifacts to the spin-echo (SE) T1-weighted andT2-weigtened or to the gradient-echo (GE) T2-weighted imag-ing sequences.

Table 1 Results of the sensor’s RR measurement capability evaluation.

Subject No. 1 2 3 4 5 6 7 8 9 10 11 12 Total

Recording time (min) 5 5 5 5 5 5 5 5 5 5 5 5 60

No. of samples (—) 72 85 49 74 62 72 89 84 66 97 84 88 924

Mean RR (rpm) 14.46 17.32 9.98 15.52 12.68 16.95 18.62 17.66 13.84 20.01 16.96 17.82 16.24

SD (rpm) 1.63 2.05 1.35 4.37 2.93 6.34 3.81 4.14 2.96 3.78 2.48 2.61 4.63

Determining distinctivepointsSamples within LoA (%) 95.83 97.65 100.00 94.59 96.77 95.95 93.26 91.67 95.45 98.97 95.24 98.86 96.10

LoA span (s) 0.17 0.15 0.19 0.15 0.18 0.17 0.15 0.16 0.15 0.17 0.16 0.16 0.16

Determining RR

Samples within LoA (%) 95.83 96.47 93.88 95.95 96.77 91.89 95.51 94.05 93.94 93.81 95.24 96.59 95.02

LoA span (rpm) 0.61 0.79 0.34 0.98 0.68 1.24 0.86 0.90 0.54 1.27 0.84 0.89 0.89

Maximum relativeerror (%)

4.40 4.88 3.69 6.79 4.95 6.70 4.74 6.41 4.16 6.12 5.08 4.97 6.83

Maximum absoluteerror (rpm)

�0.36 �0.45 �0.25 �1.17 �0.88 �1.06 �0.55 �0.89 �0.42 �0.80 �0.45 �0.48 �1.17

Fig. 10 2.5-s portion of a filtered signal with visible wave reflecting heart activity.




3.5 Signal Quality

The quality of the sensor signal depends to a large extent on thethickness of the clothing the monitored person is wearing overhis or her torso. The thicker the layer of clothing is, the moredistorted is the transmission of lung- and heart-induced vibra-tions to the sensing element. During the experimental studies,the subjects were wearing one or two layers of clothing (i.e.,a shirt or T-shirt and a sweatshirt), and no difficulties in the

physiological signal acquisition were observed. AlthoughMRI patients usually wear not more than two layers of light,indoor torso garments (room temperature), there are plans forimprovement of the system’s reliability by locating severalFBGs at various positions inside the mattress that the patientrests on.40,56 This treatment will also reduce the influence ofother factors that can deteriorate the quality of the signal,such as the sensor arrangement under the patient, the forceexerted by the lying patient on the sensor located behind

Fig. 11 Heart rate waveforms for each of the subjects in sequence extracted from

Fig. 12 Determination of (a) J peaks and (b) HR assessed using the Bland–Altman method.




Fig. 13 Relative error of HR determination as a fraction of the average value.

Table 2 Results of the sensor’s HR measurement capability evaluation.

Subject No. 1 2 3 4 5 6 7 8 9 10 11 12 Total

Recording time (min) 5 5 5 5 5 5 5 5 5 5 5 5 60

No. of samples (–) 373 268 400 407 380 334 343 325 369 335 434 366 4334

Mean HR (bpm) 71.93 50.37 81.82 83.14 75.05 64.40 67.68 63.88 73.52 64.60 82.04 73.65 74.80

SD (bpm) 3.07 2.68 4.16 6.48 5.65 6.18 3.66 4.55 3.87 4.05 4.09 5.31 4.45

Determining distinctive points


LoA span (s) 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04

Determining HR


LoA span (bpm) 3.77 1.89 4.00 3.94 3.86 2.83 3.27 2.82 3.60 3.07 5.03 3.67 3.64

Maximum relative error (%) 6.03 4.11 5.99 5.79 6.25 6.00 5.58 5.09 6.12 5.67 6.36 5.54 6.36

Maximum absoluteerror (bpm)

�2.59 �1.25 �2.57 �2.66 �2.63 �2.56 �2.17 �1.89 �2.76 �2.59 �2.96 �2.20 �2.96

Fig. 14 (a) RR and (b) HR during a complete thoracic spine MRI examination.




him, and individual features of the patient. All of these elementsdetermine the value of the signal-to-noise ratio (SNR), which forthe respiration wave presented in Fig. 6 is specified as 18.27 dBwith respect to the signal obtained during the 0.5-min period ofapnea. Similarly, the SNR level for the BCG signal shown inFig. 10 is equal to 3.37 dB, estimated on the basis of the nor-malized signal in relation to the same signal but with heart-induced disturbances removed.57

One of the main features of ballistocardiography-based sys-tems is their sensitivity to any body movements, which disturbsthe signal carrying physiological information.58 Figure 16 showsan example of a signal artifact induced by a short displacementof the track table. The artifact amplitude is five times higher thanthe signal amplitude produced by the heartbeat. Other move-ments such as coughs or changes to the body position maycause more significant distortions of the sensor signal.Therefore, BCG devices designed for clinical use usuallyhave to be equipped with additional software or hardware mod-ules that can eliminate or compensate for high-amplitude body

movements. Software filtering is the simplest means by which tocompensate, at least partially, for the disturbances. Specialarrangements of equipment that eliminate the influence ofhigh vibrations can be based, for example, on FBGs or accel-erometers not registering vibrations produced by the activity ofthe lungs or heart. The output signals in these arrangementswere used for compensating the influence of disturbances tothe sensor output signal. However, the sensor in the presentedversion can be successfully used to monitor respiration and heartrate during an MRI examination, where the patient remains stillfor tens of minutes.

4 DiscussionThe RR and HR determination capabilities of the sensor dependmainly on the signal quality, and this issue was discussed in theprevious section. In general, a noisier sensor signal indicates lessunequivocal location of the distinctive points proposed for deter-mining the RR and HR. The sensor shows maximum errors of�1.2 rpm and �3 bpm when compared with the reference RR

Fig. 15 MRI image acquired from the sensor location: (a) sagittal view, SE T2-weighted, (b) axial view, GE T2-weighted.

Fig. 16 Signal artifact caused by a short displacement of the track table.




and HR, respectively, and these values should be referred toresults obtained using other fiber-optic sensors. Although it isdifficult to compare our results with those publicized by othergroups as their outcomes are often very modest and thus notobjective, we attempt in this section to assess the proposed sen-sor with respect to other solutions.

The largest group of publications has been presented by theresearch team that worked on the OFSETH project.27–35 Theydeveloped three different fiber-optic designs, all textile-based:an FBG sensor, a macrobending sensor, and an optical timedomain reflectometry sensor. Their sensors measure elongationsof the abdominal circumference during breathing movementsand target the monitoring of sedated or anaesthetized patientsunder MRI. In their numerous publications, they provide infor-mation on the percentage textile elongation only, and make nomention of measurement error or accuracy.

Other MRI-compatible fiber-optic sensors embedded intotextiles are presented in Ref. 36, though the paper concentratesmainly on the signal processing and does not deliver quantitativedata on the sensor performance, and in Ref. 37 where the authorsverified the influence of the sensor over the quality of imaging,but did not provide sufficient respiratory data.

A simultaneous respiratory and cardiac frequency measure-ment using a single FBG sensor embedded into a polymeric foilattached to the patient’s T-shirt is described in Ref. 38. The sen-sor was tested on a group of 12 healthy subjects (ages 20 to 30years) in laboratory conditions, without exposure to the EMfield. In contrast with the results shown in this paper, wherethe figures are supplemented by the tabulated data, those authorspresent their outcomes mainly in graphical form, and therefore itis difficult to compare the two techniques objectively. However,careful analysis of the graphical data shows an improved sensi-tivity of the sensor embedded in wearable textiles; specifically,such a sensor has a BCG signal amplitude that is approximatelytwo times larger and a respiratory signal amplitude that isapproximately three times larger.

An innovative system for respiratory function monitoringcomprising an array of 40 FBGs embedded into a speciallydesigned textile vest is presented in Ref. 39. The sensor systemis distinguished by a high accuracy but is complex, and theauthors aimed at advanced measurements of three-dimensionalvolumetric changes of the human torso rather than common RRmonitoring. Another multisensor system employs 12 FBGs (in a3 × 4 array) embedded in arc-shaped metal bridges and isdeployed on the bed beneath the mattress supporting thepatient.40 The authors carried out laboratory trials on 10 subjectsand showed a maximum error of �1 rpm when compared withmanual counting. They noted the sensor’s ability to measureHR; however, corroborating data have not been made availableso far.

BCG signal acquisition has been documented in severalpapers,24–26 where a macrobending fiber-optic sensor wasembedded into a cushion to either be placed on a chair orused as a pillow. The laboratory evaluations involved five vol-unteers (ages 25 to 49 years), and the authors demonstrated anaverage accuracy above 95%; however, the maximum errorachieved was �5.5 bpm.

The list of fiber-optic devices for monitoring human vitalsigns should include interferometric and modalmetric sensors,of which the Michelson interferometer presented in Refs. 17to 19 deserves special attention as the literature provides themost relevant data. The sensor was distributed on the bed

surface and tested with the participation of 20 subjects duringrest and 10 subjects after cycling an ergometer. The measure-ment system exhibits an MRE equal to 7.35%� 7.20% forintervals between respiration events, and 3.16%� 2.32% forbeat-to-beat intervals, during rest. Slightly worse results wereobtained after physical effort. However, the maximum relativeerror exceeds the value of 30% for the RR detection and iswithin 17% for the HR detection.

Among various fiber-optic devices for monitoring vital signsin the MRI environment, the sensor presented in this paperallows respiration and heart activity signals to be acquiredsimultaneously, without the need for the patient to wear any spe-cial clothes. Compared with the results obtained by othergroups, the difference plots as well as the relative error tracesdemonstrate the RR and HR detections with an acceptable accu-racy for multiple subjects. As a logical next step following thein-depth literature analysis presented in this section, our futureworks will be directed toward multiplying sensing elements inorder to improve the system’s reliability and accuracy.

5 ConclusionsWe have proposed and tested a simple sensor design based onwell-known components of fiber-optic technology, which allowsfor noninvasive monitoring of respiration and heart activity. Themain advantage of the presented solution is its immunity to EMinterference that predisposes the sensor for use in the MRI envi-ronment. The signal from the FBG-based sensor is spectrallyencoded, enabling protection against amplitude distortions. Itwas shown that high values of the EM field generated by theMRI scanner neither affected the optical signal propagated inthe fiber nor disturbed the sensor operation. The metal-free con-struction ensured safety for the patient and did not affect thequality of imaging.

The results showed that the sensor was able to accuratelyreflect the BCG signal enabling determination of the RR andHR traces. The statistical analysis confirmed good measuringproperties of the sensor that met the basic requirements for mon-itoring a patient during an MRI examination. The Bland–Altman method allowed the depiction of the normal distributionof the data for the characteristic point determination as well as toshow the clear dependence of the data on the RR and HR valuesfor the RR and HR determination, respectively. In addition to thestatistical results, the maximum errors in RR and HR werewithin 7% (�1.2 rpm) and 6.5% (�3 bpm) of the average val-ues, respectively.

The spectrally encoded sensor signal brings another benefitin the form of easy multiplexing. This feature will enable signalacquisition from multiple FBGs located in several different pla-ces under the monitored person’s body, and addressing all theFBGs with one optical fiber. The aim of this approach will be toimprove the reliability and optimize the measuring system,while maintaining its simple design and low costs.

Future work will also concentrate on implementing the sen-sor into routine MRI examinations, which will enable increasingthe number, and diversity, of the subjects as well as performingvarious types of additional research. In particular, the subjects’behavior and their respiration rate variability will be a matter ofthorough analysis in terms of the hyperventilation phenomenonfrequently seen during MRI examinations. A deep investigationis highly desirable in the context of preventing panic symptomsin patients examined using MRI scanners and will thereforefacilitate such a diagnostic test.




AcknowledgmentsThis article includes the results of work carried out withinthe framework of Project No. DOBR/0052/R/ID1/2012/03“Development of the ORTHO-LBNP (Lower Body NegativePressure) technology for research and training of Polish AirForce pilots under conditions of ischemic hypoxia and ortho-static stress,” financed by the Program for National Defenseand Security, Poland. The Philips Achieva 1.5 T scanner waspurchased with a grant from the Fund for Polish Science andTechnology, Agreement No. 581/FNiTP/739/2010.

References1. Magnetic Resonance Imaging (MRI) Equipment—A Global Strategic

Business Report, Global Industry Analysts, Inc., San Jose, CA (2011).2. F. G. Shellock, “Patient monitoring in the MRI environment,” in

Magnetic Resonance Procedures: Health Effects and Safety, F. G.Shellock, Ed., pp. 217–241, CRC Press, Boca Raton, FL (2001).

3. A. von Smekal et al., “Patient monitoring and safety during MRI exami-nations,” Eur. Radiol. 5(3), 302–305 (1995).

4. J. C. Meléndez and E. McCrank, “Anxiety-related reactions associatedwith magnetic resonance imaging examinations,” J. Am. Med. Assoc.270(6), 745–747 (1993).

5. L. M. Harris, J. Robinson, and R. G. Menzies, “Predictors of panicsymptoms during magnetic resonance imaging scans,” Int. J. Behav.Med. 8(1), 80–87 (2001).

6. A. E. Meuret et al., “Voluntary hyperventilation in the treatment of panicdisorder - functions of hyperventilation, their implications for breathingtraining, and recommendations for standardization,” Clin. Psychol. Rev.25(3), 285–306 (2005).

7. C. L. Badour et al., “Specificity of peritraumatic fear in predicting anx-ious reactivity to a biological challenge among traumatic event-exposedadolescents,” Cogn. Ther. Res. 36(4), 397–406 (2012).

8. M. Grassi et al., “Baseline respiratory parameters in panic disorder: ameta-analysis,” J. Affect. Dis. 146(2), 158–173 (2013).

9. D. Rosenfield et al., “New breathing therapy reduces panic and anxietyby reversing hyperventilation,” J. Med. Sci. 11(8), 327–328 (2011).

10. E. Wollburg, W. T. Roth, and S. Kim, “Effects of breathing trainingon voluntary hypo- and hyperventilation in patients with panic disorderand episodic anxiety,” Appl. Psychophysiol. Biofeedback 36(2), 81–91(2011).

11. A. E. Meuret et al., “Respiratory and cognitive mediators of treatmentchange in panic disorder: evidence for intervention specificity,” J.Consult. Clin. Psychol. 78(5), 691–704 (2010).

12. A. Pittig et al., “Heart rate and heart rate variability in panic, social anxi-ety, obsessive—compulsive, and generalized anxiety disorders at base-line and in response to relaxation and hyperventilation,” Int. J.Psychophysiol. 87(1), 19–27 (2013).

13. T. Niendorf, L. Winter, and T. Frauenrath, “Electrocardiogram in anMRI environment: clinical needs, practical considerations, safetyimplications, technical solutions and future directions,” in Advancesin Electrocardiograms—Methods and Analysis, R. M. Millis, Ed.,pp. 309–324, InTech, Rijeka, Croatia (2012).

14. D. Varshneya and J. L. Maida, “Fiber optic monitor using interferometryfor detecting vital signs of a patient,” U.S. Patent No. 6498652 (2002).

15. M. Życzkowski et al., “Interferometric fiber optics based sensor for mon-itoring of the heart activity,” Acta Phys. Pol. A 120(4), 782–784 (2011).

16. F. C. Favero, J. Villatoro, and V. Pruneri, “Microstructured optical fiberinterferometric breathing sensor,” J. Biomed. Opt. 17(3), 037006 (2012).

17. D. Zazula, D.Đonlagic, and S. Šprager, “Application of fibre-optic inter-ferometry to detection of human vital signs,” J. Laser Health Acad. 1,27–32 (2012).

18. S. Šprager and D. Zazula, “Heartbeat and respiration detection fromoptical interferometric signals by using a multimethod approach,”IEEE Trans. Biomed. Eng. 59(10), 2922–2929 (2012).

19. S. Šprager and D. Zazula, “Detection of heartbeat and respiration fromoptical interferometric signal by using wavelet transform,” Comput.Methods Programs Biomed. (2013) in press.

20. M. Zyczkowski et al., “Using modalmetric fiber optic sensors to mon-itor the activity of the heart,” Proc. SPIE 7894, 789404 (2011).

21. L. Dziuda et al., “The device for measuring vital signs,” Patent appli-cation No. P 391763, Patent Office of the Republic of Poland (2010).

22. L. Dziuda et al., “Fiber-optic sensor for monitoring respiration and car-diac activity,” in IEEE Sens. Conf. Proc., Limerick, Ireland, pp. 413–416 (2011).

23. L. Dziuda et al., “Monitoring respiration and cardiac activity using fiberBragg grating-based sensor,” IEEE Trans. Biomed. Eng. 59(7), 1934–1942 (2012).

24. Z. Chen et al., “Smart pillow for heart-rate monitoring using a fiber opticsensor,” Proc. SPIE 7894, 789402 (2011).

25. Z. Chen et al., “Portable fiber optic ballistocardiogram sensor for homeuse,” Proc. SPIE 8818, 82180X (2012).

26. C. J. Deepu et al., “A smart cushion for real-time heart rate monitoring,”in IEEE BioCAS Conf. Proc., Hsinchu, Taiwan, pp. 53–56, IEEE(2012).

27. J. De jonckheere, “OFSETH: Optical fibre embedded into technical tex-tile for healthcare, an efficiency way to monitor patient under magneticresonance imaging,” in IEEE EMBS Conf. Proc., Lyon, France,pp. 3950–3953 (2007).

28. L. T. D'Angelo et al., “A system for respiratory motion detection usingoptical fibers embedded into textiles,” in IEEE EMBS Conf. Proc.,Vancouver, Canada, pp. 3694–3697 (2008).

29. J. De jonckheere et al., “FBG-based smart textiles for continuous mon-itoring of respiratory movements for healthcare applications,” in IEEEHealthcom Conf. Proc., Lyon, France, pp. 277–282 (2010).

30. J. De jonckheere et al., “Optical fibre sensors embedded into technicaltextile for a continuous monitoring of patients under magnetic reso-nance imaging,” in IEEE EMBS Conf. Proc., Vancouver, Canada,pp. 5266–5269 (2008).

31. F. Narbonneau et al., “OFSETH: Optical technologies embedded insmart medical textile for continuous monitoring of respiratory motionsunder magnetic resonance imaging,” Proc. SPIE 7715, 7715D-1 (2010).

32. J. De jonckheere et al., “OFSETH: a breathing motions monitoring sys-tem for patients under MRI,” in IEEE EMBS Conf. Proc., Buenos Aires,Argentina, pp. 1016–1019 (2010).

33. K. Krebber et al., “Smart technical textiles with integrated POF sen-sors,” Proc. SPIE 6933, 69330V (2008).

34. A. Grillet et al., “Optical fiber sensors embedded into medical textilesfor healthcare monitoring,” IEEE Sens. J. 8(7), 1215–1222 (2008).

35. J. Witt et al., “Smart medical textiles with embedded optical fibre sen-sors for continuous monitoring of respiratory movements during MRI,”Proc. SPIE 7653, 76533B (2010).

36. C. Zhang et al., “Smart textile sensing system for human respirationmonitoring based on fiber Bragg grating,” Proc. SPIE 7381, 738104(2009).

37. W. Jae Yoo et al., “Development of respiration sensors using plasticoptical fiber for respiratory monitoring inside MRI system,” J. Opt.Soc. Korea 14(3), 235–239 (2010).

38. A. F. Silva et al., “Simultaneous cardiac and respiratory frequency meas-urement based on a single fiber Bragg grating sensor,” Meas. Sci.Technol. 22(7), 075801 (2011).

39. T. Allsop et al., “Respiratory function monitoring using a real-timethree-dimensional fiber-optic shaping sensing scheme based uponfiber Bragg gratings,” J. Biomed. Opt. 17(11), 117001 (2012).

40. J. Hao et al., “FBG-based smart bed system for healthcare applications,”Front. Optoelectron. China 3(1), 78–83 (2010).

41. M. Majumder et al., “Fibre Bragg gratings in structural health monitor-ing—present status and applications,” Sensor. Actuator. A Phys. 147(1),150–164 (2008).

42. L. Dziuda et al., “Hybrid fiber optic voltage sensor for remote monitoringof electrical submersible pump motors,” Opt. Eng. 44(6), 064401 (2005).

43. E. Al-Fakih et al., “The use of fiber Bragg grating sensors in biomechan-ics and rehabilitation applications: the state-of-the-art and ongoingresearch topics,” Sensors 12(10), 12890–12926 (2012).

44. R. A. Betts et al., “Nonlinear refractive index in erbium doped opticalfiber: theory and experiment,” IEEE J. Quant. Electron. 27(4), 908–913(1991).

45. Micron Optics, Inc., “Optical sensing interrogator sm130,” MicronOptics Technical Information, http://www.micronoptics.com/ (2010).

46. Invivo International, “Precess (model 3160) MRI patient monitoringsystem,” Invivo Technical Data Sheet, [Online], http://www.invivocorp.com/ (2010).




http://dx.doi.org/10.1001/jama.1993.03510060091039

http://dx.doi.org/10.1016/j.cpr.2005.01.002

http://dx.doi.org/10.1007/s10608-011-9380-0

http://dx.doi.org/10.1016/j.jad.2012.08.034

http://dx.doi.org/10.1007/s10484-011-9150-5

http://dx.doi.org/10.1037/a0019552

http://dx.doi.org/10.1037/a0019552

http://dx.doi.org/10.1016/j.ijpsycho.2012.10.012

http://dx.doi.org/10.1016/j.ijpsycho.2012.10.012


http://dx.doi.org/10.1109/TBME.2012.2213302

http://dx.doi.org/10.1016/j.cmpb.2013.03.003

http://dx.doi.org/10.1016/j.cmpb.2013.03.003

http://dx.doi.org/10.1117/12.873379


http://dx.doi.org/10.1117/12.871208

http://dx.doi.org/10.1117/12.909768

http://dx.doi.org/10.1117/12.854386

http://dx.doi.org/10.1117/12.776758

http://dx.doi.org/10.1109/JSEN.2008.926518

http://dx.doi.org/10.1117/12.866434

http://dx.doi.org/10.1117/12.834970

http://dx.doi.org/10.3807/JOSK.2010.14.3.235

http://dx.doi.org/10.3807/JOSK.2010.14.3.235

http://dx.doi.org/10.1088/0957-0233/22/7/075801

http://dx.doi.org/10.1088/0957-0233/22/7/075801


http://dx.doi.org/10.1007/s12200-009-0066-0

http://dx.doi.org/10.1016/j.sna.2008.04.008

http://dx.doi.org/10.1117/1.1926947

http://dx.doi.org/10.3390/s121012890

http://dx.doi.org/10.1109/3.83325

http://www.micronoptics.com/



http://www.invivocorp.com/



47. K. Youngho et al., “Evaluation of unconstrained monitoring technologyused in the smart bed for u-Health environment,” Telemed. J. e-Health17(6), 435–441 (2011).

48. J. M. Bland and D. G. Altman, “Statistical methods for assessing agree-ment between two methods of clinical measurement,” Lancet327(8476), 307–310 (1986).

49. J. M. Bland and D. G. Altman, “Measuring agreement in method com-parison studies,” Stat. Meth. Med. Res. 8(2), 135–160 (1999).

50. R. S. Gubner, M. Rodstein, and H. E. Ungerleider, “Ballistocardiogra-phy: an appraisal of technic, physiologic principles, and clinical value,”Circulation 7, 268–286 (1953).

51. O. T. Inan et al., “Robust ballistocardiogram acquisition for home mon-itoring,” Phys. Meas. 30(2), 169–185 (2009).

52. O. T. Inan et al., “Noninvasive measurement of physiological signals ona modified home bathroom scale,” IEEE Trans. Biomed. Eng. 59(8),2137–2143 (2012).

53. L. Giovangrandi et al., “Ballistocardiography–a method worth revisit-ing,” in IEEE EMBS Conf. Proc., Boston, pp. 4279–4282, IEEE (2011).

54. W. H. Bancroft, Jr., “Kinetocardiography: past and present,” Bibl.Cardiol. 37, 58–72 (1979).

55. R. Maniewski, Magnetic Examinations of Electrical and MechanicalCardiac Activity, Ossolineum, Wroclaw (1987).

56. M. Nishyama, M. Miyamoto, and K. Watanabe, “Respiration and bodymovement analysis during sleep in bed using hetero-core fiber opticpressure sensors without constraint to human activity,” J. Biomed.Opt. 16(1), 017002 (2011).

57. O. T. Inan, G. T. Kovacs, and L. Giovangrandi, “Evaluating the lower-body electromyogram signal acquired from the feet as a noise referencefor standing ballistocardiogram measurements,” IEEE Trans. Inform.Technol. Biomed. 14(5), 1188–1196 (2010).

58. R. M. Wiard et al., “Automatic detection of motion artifacts in theballistocardiogram measured on a modified bathroom scale,” Med.Biol. Eng. Comput. 49(2), 213–220 (2011).




http://dx.doi.org/10.1089/tmj.2010.0211

http://dx.doi.org/10.1016/S0140-6736(86)90837-8

http://dx.doi.org/10.1191/096228099673819272

http://dx.doi.org/10.1161/01.CIR.7.2.268

http://dx.doi.org/10.1088/0967-3334/30/2/005


http://dx.doi.org/10.1117/1.3528008

http://dx.doi.org/10.1117/1.3528008

http://dx.doi.org/10.1109/TITB.2010.2044185

http://dx.doi.org/10.1109/TITB.2010.2044185

http://dx.doi.org/10.1007/s11517-010-0722-y

http://dx.doi.org/10.1007/s11517-010-0722-y

Documents

Fiber Bragg grating-based sensor for monitoring ... · Fiber Bragg grating-based sensor for monitoring respiration and heart activity during magnetic resonance imaging examinations